New Polls Flip Portugal’s Presidential Race and Fuel Voter Doubts

A fresh poll released this week did more than reshuffle the ranking of Portugal’s presidential hopefuls—it reignited a fierce row over how surveys are carried out and why their results diverge so sharply. In cafés, on radio call-ins and across social media, voters are asking the same question: can any of these numbers be trusted as the January ballot approaches?

A survey landscape in turmoil

The newest results, commissioned by a Lisbon-based media consortium, place Luís Marques Mendes marginally ahead of Henrique Gouveia e Melo and André Ventura, yet only days earlier a rival study crowned Ventura the frontrunner. A third institute, meanwhile, still shows the former navy chief tied for first. The gulf between the figures is striking because each poll was taken within November’s narrow window, all before many viewers had even tuned in to the first televised debate. Statistically, the gaps fall inside the declared margin of error, but the optics are anything but tidy. For electorate-watchers in Portugal, such wild oscillations conjure memories of the 2021 municipal races when projections veered off course in the campaign’s final stretch. The common thread today is the near certainty of a second-round runoff, an outcome hinted at by every survey, however contradictory the leader board may look.

What the numbers show — and hide

On paper, the most recent Pitagórica exercise has Mendes securing roughly one quarter of the vote, with Gouveia e Melo and Ventura separated by mere decimals. A simultaneous ICS/ISCTE study published in a Sunday weekly places Gouveia e Melo fractionally in front after undecided voters are allocated. In that same table, João Cotrim de Figueiredo languishes at 3 %, a collapse he swiftly labelled “incredible” and “potentially manipulative”. Two weeks before, an Intercampus snapshot had awarded him more than triple that share. Analysts note that small tweaks in how the undecided bloc—still hovering near 20 %—is distributed can catapult or sink a candidacy overnight. Equally crucial is the timing: the ICS/ISCTE fieldwork ended before the duel on RTP where Cotrim de Figueiredo scored a widely praised exchange with Ventura. In other words, voters consulted for that poll never saw the debate many pundits believe re-energised the liberal contender.

Methodology under the microscope

Behind every headline figure sits a chain of methodological choices. Sample size varies from 800 interviews in the ICS/ISCTE exercise to 1 000 in the Pitagórica survey, while the response rate in phone-based polls now hovers near 1 %, a far cry from the 60 % of the 1980s. Under such constraints, statisticians warn that weighting models—meant to correct for the under-representation of young or rural voters—can introduce fresh distortions. The treatment of indecisive respondents, routinely re-allocated according to age or past turnout, differs markedly from one institute to another. Even the order of candidate names read to participants can nudge outcomes by a point or two. Regulatory agencies such as the ERC and the CNE require a technical appendix, yet few citizens parse those PDFs, leaving ample room for political actors to cherry-pick whichever result flatters their narrative.

Political reactions and public perception

Mendes’s headquarters quickly hailed the newest chart as proof of “growing momentum”. Gouveia e Melo’s strategists counter that their man still wins every simulated runoff scenario. Over at Chega, Ventura rails against “urban-elite polling houses” whenever a graph shows him slipping, only to share the very next one should it place him on top. Cotrim de Figueiredo has escalated matters by filing a formal complaint to the media regulator, alleging that one particular survey “may disorient the electorate”. Outside campaign war rooms, veteran poll-watchers like professor Sónia Nunes of ISCTE remind observers that Portuguese presidential races have historically delivered surprises: in 2006, final-week swings surpassed 8 % for some candidates, well within today’s most pessimistic confidence intervals. Still, trust is fragile. A study by the University of Porto last year found that only 42 % of Portuguese adults believe polling results “reflect reality most of the time”.

Reading the tea leaves ahead of January

With the first round set for 18 January 2026 and televised debates running until Christmas, the polling chessboard will almost certainly shift again. Experts advise voters to focus less on precise ranks and more on broader trends: whether a name is rising, stalling or falling for several waves in a row. They also urge scrutiny of the field-work dates, sample composition and whether percentages refer to raw responses or post-distribution figures. In the end, the most reliable poll remains election day itself, a fact underlined by the parliamentary contests of 2022 when final pre-vote averages misjudged Chega’s surge by nearly four points. Until the ballots are counted, Portugal’s electorate would do well to treat every new graph as a snapshot rather than a prophecy, however loudly the campaigns trumpet—or denounce—each successive curve.